The invention generally relates to an automobile accident detection and data recordation and reporting system, and in particular to a system which detects accidents based on a set of characteristic sounds or other cues.
Traffic accidents cause significant costs in terms of direct loss, consequential loss, and societal loss due to obstruction of the roadway in the aftermath of an accident. Another issue is the allocation of direct costs, for example when more than one vehicle is involved, the vehicle at fault is generally held liable for the damages.
It is possible to monitor locations that are likely places for accidents to occur, however, without intelligence, this process may be inefficient and unproductive. Likewise, without immediate and efficient communication of the information obtained, benefits of the monitoring are quite limited.
Since cellular telephone technology has become so widely adopted, the most common means by which motor vehicle accidents are reported to agencies in the U.S. is through cellular telephones. However, this is not always reliable or immediate if the victims are unable to use their cellular phones or if there are no witnesses with cellular phones to report the accident, and it fails to record an actual record of the accident which can later be used as evidence.
Automobile accident detection systems are common in the art. Upon the occurrence of an automobile accident, it may be desirable to obtain video images and sounds of the accident and to record the time of the accident and the status of the traffic lights at the time the accident occurred. This information can then be sent to a remote location where emergency crews can be dispatched and the information further examined and forwarded to authorities in order to determine fault and liability.
A number of prior art techniques are available for predicting the occurrence of an accident. Some of these require an extended period of time for an automated system to analyze the data, and thus any report generated is substantially delayed. In others, the accuracy of the system depends on environmental conditions, such as lighting or time of day. Therefore, in order to provide an immediate and reliable response to a predicted occurrence of an accident, such techniques are suboptimal.
For example, Japanese Patent Application No. 8-162911 entitled “Motor Vehicle Accident Monitoring Device” (“the Japanese reference”), expressly incorporated herein by reference in its entirety, discloses a system for monitoring traffic accidents including a plurality of microphones and video cameras disposed at an intersection. Collision sounds are chosen from among the typical sounds at an intersection. The source of the collision sounds is determined by comparing the time differences of the sounds received by each of the microphones. Image data from the cameras is recorded upon the occurrence of the collision. However, the Japanese reference discloses a system that is constantly photographing the accident scene thereby wasting video resources.
U.S. Pat. No. 6,141,611 issued to Mackey et al. entitled “Mobile Vehicle Accident Data System” (“the Mackey reference”), expressly incorporated herein by reference in its entirety, discloses an on-board vehicle accident detection system including one or more video cameras that continuously record events occurring at a given scene. Camera images of the scene are digitally stored after compression. An accident detector on-board the vehicle determines if an accident has occurred, and if so, the stored images are transmitted to a remote site for observation. However, the Mackey reference includes video cameras on-board the vehicles themselves, increasing the likelihood that the cameras would become damaged during an accident thereby rendering them impractical for accident-recording systems. Further, the on-board cameras' image-capturing ability is severely limited due to the constraints of the vehicle themselves. Additionally, the Mackey reference discloses a system that determines if an accident is present by the sudden acceleration or deceleration of the vehicle, without the use of fixed microphones. The invention claimed by Mackey is on board the vehicle, it does nothing to solve the problem or record an accident in two vehicles which are not so equipped. Equipping every vehicle with this system is impractical and therefore not feasible.
U.S. Pat. No. 6,111,523 issued to Mee entitled “Method and Apparatus for Photographing Traffic in an Intersection”, expressly incorporated herein by reference in its entirety, describes a system for taking photographs of vehicles at a traffic intersection by triggering a video camera to capture images wherein the triggering mechanism of the video camera is based upon certain vehicle parameters including the speed of the vehicle prior to its entrance into the traffic intersection.
U.S. Pat. No. 6,088,635 issued to Cox et al. entitled “Railroad Vehicle Accident Video Recorder”, expressly incorporated herein by reference in its entirety, discloses a system for monitoring the status of a railroad vehicle prior to a potential accident. The system employs a video camera mounted within the railroad car that continuously views the status of a given scene, and continuously stores the images of the scene. Like Mackey, it is impractical and therefore not feasible to equip every vehicle with this system.
U.S. Pat. No. 5,717,391 issued to Rodriguez entitled “Traffic Event Recording Method and Apparatus”, expressly incorporated herein by reference in its entirety, describes a system for determining the condition of a traffic light and includes an audio sensor which monitors sound at all times. Sound detected above a certain decibel level triggers the recordation of sounds, the time of day and the status of the traffic lights. However, Rodriguez fails to disclose video cameras or any image-capturing means.
U.S. Pat. No. 5,677,684 issued to McArthur entitled “Emergency Vehicle Sound-Actuated Traffic Controller”, expressly incorporated herein by reference in its entirety, describes a traffic controller system utilizing sound detection means connected to a control box which contains a switching mechanism that, in a first orientation, allows normal operation of traffic light control and a second orientation that, upon the detection of an approaching siren, sets all traffic signals at an intersection to red to prohibit the entrance into the intersection of additional vehicles.
U.S. Pat. No. 5,539,398 issued to Hall et al. entitled “GPS-based Traffic Control Preemption System”, expressly incorporated herein by reference in its entirety, discloses a system for determining if a vehicle issuing a preemption request to an emergency vehicle or police car is within an allowed approach of a traffic intersection, utilizing a GPS system.
U.S. Pat. No. 6,690,294 issued to Zierden entitled “System and method for detecting and identifying traffic law violators and issuing citations”, expressly incorporated herein by reference, discloses a mobile or stationary traffic monitoring system for detecting violations of speed limits or other traffic laws by vehicle operators and issuing citations to an operator and/or vehicle owner suspected of a violation using a digital camera to capture images of the operator and/or the vehicle, transmitting the captured images and other relevant data to an analysis center where the images and data are analyzed to determine whether to issue a citation and, if so, to issue the citation or take other appropriate law enforcement measures. The system captures images of a vehicle and/or vehicle operator suspected of a traffic violation, determines the time and geographic location of the suspected violation, transmits the images and other data to an analysis center, issues citations to violators and derives revenue therefrom.
U.S. Pat. No. 5,938,717 to Dunne et al., expressly incorporated herein by reference, discloses a traffic control system that automatically captures an image of a vehicle and speed information associated with the vehicle and stores the image and information on a hard disk drive. The system uses a laser gun to determine whether a vehicle is speeding. The hard drive is later connected to a base station computer which is, in turn, connected to a LAN at which the information from the hard drive is compared with databases containing data such as vehicle registration information and the like. The system automatically prints a speeding citation and an envelope for mailing to the registered owner of the vehicle
U.S. Pat. No. 5,734,337 to Kupersmit, expressly incorporated herein by reference, discloses a stationary traffic control method and system for determining the speed of a vehicle by generating two images of a moving vehicle and calculating the vehicle speed by determining the distance traveled by the vehicle and the time interval between the two images. The system is capable of automatically looking up vehicle ownership information and issuing citations to the owner of a vehicle determined to be speeding.
U.S. Pat. No. 5,948,038 to Daly et al., expressly incorporated herein by reference, discloses a method for processing traffic violation citations. The method includes the steps of determining whether a vehicle is violating a traffic law, recording an image of the vehicle committing the violation, recording deployment data corresponding to the violation, matching the vehicle information with vehicle registration information to identify the owner, and providing a traffic violation citation with an image of the vehicle, and the identity of the registered owner of the vehicle.
The I-95 Corridor Coalition, Surveillance Requirements/Technology, Ch. 4., Technology Assessment, expressly incorporated herein by reference, describes a number of different technologies suitable for incident detection. For example, AutoAlert: Automated Acoustic Detection of Traffic Incidents, was an IVHS-IDEA project which uses military acoustic sensor technologies, e.g., AT&T IVHS NET-2000™. The AutoAlert system monitors background traffic noise and compares it with the acoustic signatures of previously recorded accidents and incidents for detection. See, David A. Whitney and Joseph J. Pisano (TASC, Inc., Reading, Mass.), “AutoAlert: Automated Acoustic Detection of Incidents”, IDEA Project Final Report, Contract ITS-19, IDEA Program, Transportation Research Board, National Research Council, Dec. 26, 1995, expressly incorporated herein by reference. The AutoAlert system employs algorithms which provide rapid incident detection and high reliability by applying statistical models, including Hidden Markov Models (HMM) and Canonical Variates Analysis (CVA). These are used to analyze both short-term and time-varying signals that characterize incidents.
The Smart Call Box project (in San Diego, Calif.) evaluated the use of the existing motorist aid call box system for other traffic management strategies. The system tests the conversion of existing cellular-based call boxes to multifunctional IVHS system components, to transmit the data necessary for traffic monitoring, incident detection, hazardous weather detection, changeable message sign control, and CCTV control.
In 1992 the French Toll Motorway Companies Union initiated testing an Automatic Incident Detection (AID) technique proposed by the French National Institute for Research on Transportation and Security (INRETS). The technique consists of utilizing computers to analyze video images received by television cameras placed along the roadway. A “mask” frames the significant part of the image, which typically is a three or four-lane roadway and the emergency shoulder. The computer processes five pictures a second, compares them two at a time, and analyzes them looking for points that have moved between two successive pictures. These points are treated as objects moving along the roadway. If a moving object stops and remains stopped within the mask for over 15 seconds, the computer considers this an anomaly and sets off an alarm. In 1993, as part of the European MELYSSA project, the AREA Company conducted a full scale test over an urban section of the A43 motorway located east of Lyons. The roadway was equipped with 16 cameras on 10 meter masts or bridges with focal distances varying from 16 to 100 km, and fields of detection oscillating between 150 and 600 meters. Image Processing and Automatic Computer Traffic Surveillance (IMPACTS) is a computer system for automatic traffic surveillance and incident detection using output from CCTV cameras. The algorithm utilized by the IMPACTS system takes a different approach from most other image processing techniques that have been applied to traffic monitoring. Road space and how it is being utilized by traffic is considered instead of identifying individual vehicles. This leads to a qualitative description of how the road, within a CCTV image, is occupied in terms of regions of empty road or moving or stationary traffic. The Paris London Evaluation of Integrated ATT and DRIVE Experimental Systems (PLEIADES) is part of the DRIVE Research Programme. The Automatic Traffic Surveillance (ATS) system has been installed into Maidstone Traffic Control Center and provides information on four separate CCTV images. This information will be used both in the Control Center and passed onto the Traffic Information Center via the PLEIADES Information Controller (PIC) and data communications link. Instead of remote PCs there is a duplicate display of the Engineers workstation that is shown in the Control Office on a single computer monitor. The ATS system communicates data at regular intervals to the PIC. Any alarms that get raised or cleared during normal processing will get communicated to the PIC as they occur. The PIC uses the information received to display a concise picture of a variety of information about the highway region. The ATS system uses video from CCTV cameras taken from the existing Control Office Camera Multiplex matrix, while not interfering with its normal operation. When a camera is taken under manual control, the processing of the data for that image is suspended until the camera is returned to its preset position.
Navaneethakrishnan Balraj, “Automated Accident Detection In Intersections Via Digital Audio Signal Processing” (Thesis, Mississippi State University, December 2003), expressly incorporated herein by reference, discusses, inter alia, feature extraction from audio signals for accident detection. The basic idea of feature extraction is to represent the important and unique characteristics of each signal in the form of a feature vector, which can be further classified as crash or non-crash using a statistical classifier or a neural network. Others have tried using wavelet and cepstral transforms to extract features from audio signals such as speech signals. S. Kadambe, G. F. Boudreaux-Bartels, “Application of the wavelet transform for pitch detection of speech signals,” IEEE Trans. on Information Theory, vol. 38, no. 2, part 2, pp. 917-924, 1992; C. Harlow and Y. Wang, “Automated Accident Detection,” Proc. Transportation Research Board 80th Annual Meeting, pp 90-93, 2001. Kadambe et al developed a pitch detector using a wavelet transform. One of the main properties of the dyadic wavelet transform is that it is linear and shift-variant. Another important property of the dyadic wavelet transform is that its coefficients have local maxima at a particular time when the signal has sharp changes or discontinuities. These two important properties of the dyadic wavelet transform help to extract the unique features of a particular audio signal. Kadambe et al made a comparison of the results obtained from using dyadic wavelet transforms, autocorrelation, and cepstral transforms. The investigation showed that the dyadic wavelet transform pitch detector gave 100% accurate results. One reason for the difference in the results was that the other two methods assume stationarity within the signal and measure the average period, where as the dyadic wavelet transform takes into account the non-stationarities in the signal. Hence, the dyadic wavelet transform method would be the best to extract feature when the signals are non-stationary. Harlow et al developed an algorithm to detect traffic accidents at intersections, using an audio signal as the input to the system. The algorithm uses the Real Cepstral Transform (RCT) as a method to extract features. The signals recorded at intersections include brake, pile drive, construction and normal traffic sounds. These signals are segmented into three-second sections. Each of these three second segmented signals is analyzed using RCT. RCT is a method where the signal is windowed for every 100 msec using a hamming window with an overlap of 50 msec. Thus, for a given three-second signal, there will be almost 60 segments of 100 msec duration each. RCT is applied to each of these segments, and the first 12 coefficients are used as the features. The features obtained using the RCT are then classified as “crash” or “non-crash” using a neural network.
Balraj's experimental results showed that among the three different statistical classifiers investigated, maximum likelihood and nearest neighbor performed best, although this had high computational costs. Haar, Daubechies, and Coiflets provided the best classification accuracies for a two-class system. Among the five different feature extraction methods analyzed on the basis of the overall accuracy, RCT performed best. The second-generation wavelet method, the lifting scheme, was also investigated. It proved computationally efficient when compared to DWT. Thus, it was concluded that the optimum design for an automated system would be a wavelet-based feature extractor with a maximum likelihood classifier. Thus the choice of DWT or the lifting scheme would be preferred for a real-time system.
In any and/or all of the embodiments described herein, the systems, equipment systems, subsystems, devices, components, and/or appliances, of and/or utilized in any of the respective embodiments, can include and/or can utilize the teachings and/or the subject matter of the following U.S. Patents, the subject matter and teachings of which are hereby incorporated by reference herein and form a part of the disclosure of this patent application: U.S. Pat. No. 6,009,356 (Monroe, Dec. 28, 1999); U.S. Pat. No. 5,890,079 (Beemer, II, et al., Sep. 7, 1999); U.S. Pat. No. 5,845,240 (Fielder, Dec. 1, 1998); U.S. Pat. No. 5,948,026 (Levine, Mar. 30, 1999); U.S. Pat. No. 5,446,659 (Yamawaki, Aug. 29, 1995); U.S. Pat. No. 5,056,056 (Gustin, Oct. 8, 1991); U.S. Pat. No. 6,718,239 (Rayner, Apr. 6, 2004); U.S. Pat. No. 6,449,540 (Rayner, Sep. 10, 2002); U.S. Pat. No. 6,684,137 (Takagi, et al., Jan. 27, 2004); U.S. Pat. No. 6,633,238 (Lemelson, et al., Oct. 14, 2003); U.S. Pat. No. 6,226,389 (Lemelson, et al., May 1, 2001); U.S. Pat. No. 6,630,884 (Shanmugham, Oct. 7, 2003); U.S. Pat. No. 6,600,417 (Lerg, et al., Jul. 29, 2003); U.S. Pat. No. 6,288,643 (Lerg, et al., Sep. 11, 2001); U.S. Pat. No. 6,281,792 (Lerg, et al., Aug. 28, 2001); U.S. Pat. No. 6,580,373 (Ohashi, Jun. 17, 2003); U.S. Pat. No. 6,574,548 (DeKock, et al., Jun. 3, 2003); U.S. Pat. No. 6,472,982 (Eida, et al., Oct. 29, 2002); U.S. Pat. No. 5,784,007 (Pepper, Jul. 21, 1998); U.S. Pat. No. 6,466,260 (Hatae, et al., Oct. 15, 2002); U.S. Pat. No. 6,429,812 (Hoffberg, Aug. 6, 2002); U.S. Pat. No. 6,252,544 (Hoffberg, Jun. 26, 2001); U.S. Pat. No. 6,389,340 (Rayner, May 14, 2002); U.S. Pat. No. 6,324,450 (Iwama, Nov. 27, 2001); U.S. Pat. No. 6,211,907 (Scaman, et al., Apr. 3, 2001); and U.S. Pat. No. 5,689,442 (Swanson, et al., Nov. 18, 1997); U.S. Pat. No. 6,647,270 (Himmelstein, Nov. 11, 2003); U.S. Pat. No. 6,574,538 (Sasaki, Jun. 3, 2003); U.S. Pat. No. 6,573,831 (Ikeda, et al., Jun. 3, 2003); U.S. Pat. No. 6,542,077 (Joao, Apr. 1, 2003); U.S. Pat. No. 6,404,352 (Ichikawa, et al., Jun. 11, 2002); U.S. Pat. No. 6,401,027 (Xu, et al., Jun. 4, 2002); U.S. Pat. No. 6,392,692 (Monroe, May 21, 2002); U.S. Pat. No. 6,339,370 (Ruhl, et al., Jan. 15, 2002); U.S. Pat. No. 6,314,364 (Nakamura, Nov. 6, 2001); U.S. Pat. No. 6,163,338 (Johnson, et al., Dec. 19, 2000); U.S. Pat. No. 6,154,658 (Caci, Nov. 28, 2000); U.S. Pat. No. 6,091,956 (Hollenberg, Jul. 18, 2000); U.S. Pat. No. 6,087,960 (Kyouno, et al., Jul. 11, 2000); U.S. Pat. No. 6,075,466 (Cohen, et al., Jun. 13, 2000); U.S. Pat. No. 5,990,801 (Kyouno, et al., Nov. 23, 1999); U.S. Pat. No. 5,943,428 (Seri, et al., Aug. 24, 1999); U.S. Pat. No. 5,699,056 (Yoshida, Dec. 16, 1997); U.S. Pat. No. 5,353,023 (Mitsugi, Oct. 4, 1994); U.S. Pat. No. 5,025,324 (Hashimoto, Jun. 18, 1991); U.S. 20040022416 (Lemelson, et al., Feb. 5, 2004); U.S. 20020008637 (Lemelson, et al., Jan. 24, 2002); U.S. 20030225516 (DeKock, et al., Dec. 4, 2003); U.S. 20020193938 (DeKock, et al., Dec. 19, 2002); U.S. 20030222981 (Kisak, et al., Dec. 4, 2003); U.S. 20030214405 (Lerg, et al., Nov. 20, 2003); U.S. 20020008619 (Lerg, et al., Jan. 24, 2002); U.S. 20030125853 (Takagi, et al., Jul. 3, 2003); U.S. 20030081935, 20030081934, 20030081128, 20030081127, 20030081122, 20030081121, and 20030080878 (Kirmuss, May 1, 2003); U.S. 20020121969 (Joao, Sep. 5, 2002); U.S. 20020147982 (Naidoo, et al., Oct. 10, 2002); U.S. 20030062997 (Naidoo, et al., Apr. 3, 2003); U.S. 20010005804 (Rayner, Jun. 28, 2001); U.S. 20020163579 (Patel, et al., Nov. 7, 2002); U.S. 20020170685 (Weik, et al., Nov. 21, 2002); U.S. 20030011684 (Narayanaswami, et al., Jan. 16, 2003); U.S. 20030041329 (Bassett, Feb. 27, 2003); U.S. 20030053536 (Ebrami, Mar. 20, 2003); and U.S. 20030067542 (Monroe, Apr. 10, 2003).
The following references are incorporated herein by references as if explicitly set forth herein:
www.itsdocs.fhwa.dot.gov/JPODOCS/REPTS_TE/36D01!.PDF;
www.ndsu.nodak.edu/ndsu/ugpti/MPC_Pubs/html/MPC01-122.html;
www.intelligenthighway.com/ITS/IMITS.pdf,
stat-www.berkeley.edu/users/kwon/papers/inc_detection.pdf;
www-users.cs.umn.edu/˜masoud/publications/harini-intersection-itsc-2002.pdf.,
The I-80 Experiment: Real-Time Algorithms for Travel Time Estimates and Incident Detection—Alex Skabardonis;
Dougherty, M. S., Chen, H. and Montgomery, F. (1998), Principle components based incident detection Proc. 5th International Conference on Applications of Advanced Technologies in Transportation, Newport Beach, Calif., ieeexplore.ieee.org/xpl/tocresult.jsp?isNumber=14013;
Karim, A. and Adeli, H. (2003), “Fast Automatic Incident Detection on Urban and Rural Freeways Using the Wavelet Energy Algorithm,” Journal of Transportation Engineering, ASCE, Vol. 129, No. 1, pp. 57-68. www.pubs.asce.org/WWWdisplay.cgi?0300116;
Chen, H; Boyle, R D; Montgomery, F O; Kirby, H R. Novelty detection for motorway incident management in: Proceedings of the Workshop on Engineering for Intelligent Systems. 1998. portal.acm.org/citation.cfm?id=643217&dl=ACM&coll=GUIDE;
Karim, A. and Adeli, H. (2002), “Comparison of Fuzzy Wavelet Radial Basis Function Neural Network Freeway Incident Detection Model with California Algorithm,” Journal of Transportation Engineering, ASCE, Vol. 128, No. 1, pp. 21-30, www.pubs.asce.org/WWWdisplay.cgi?0200117;
Chien-Hua Hsiao, Ching-Teng Lin, and Michael Cassidy, “Application of Fuzzy Logic and Neural Networks to Automatically Detect Freeway Traffic Incidents”, Journal of Transportation Engineering, Vol. 120, No. 5, September/October 1994, pp. 753-772;
A. Samant & H. Adeli, “Feature Extraction for Traffic Incident Detection Using Wavelet Transform and Linear Discriminant Analysis”, Computer-Aided Civil and Infrastructure Engineering, Volume 15 Issue 4 Page 241 (July 2000);
Hojjat Adelil and Asim Karim, “Fuzzy-Wavelet RBFFN Model For Freeway Incident Detection”, Journal of Transportation Engineering, Vol. 126, No. 6, November/December, 2000;
H. Veeraraghavan, O. Masoud, N. P. Papanikolopoulos, “Computer vision algorithms for intersection monitoring”, IEEE Trans. on Intelligent Transportation Systems, vol. 4, no. 2, pp. 78-89, June 2003;
K. Stubbs, H. Arumugam, O. Masoud, C. McMillen, H. Veeraraghavan, R. Janardan, N. P. Papanikolopoulos, “A real-time collision warning system for intersections”, in Proc. ITS America 13th Annual Meeting, Minneapolis, Minn., May 2003.;
H. Veeraraghavan, O. Masoud, N. P., Papanikolopoulos, “Vision-based monitoring of intersections”, in Proc. IEEE 5th International Conference on Intelligent Transportation Systems, pp. 7-12, Singapore, September 2002.;
Petty, K., Ostland, M., Kwon, J., Rice, J. and Bickel, P. (2002) “A New Methodology for Evaluating Incident Detection Algorithms,” Transportation Research, Part C, Volume 10, pp. 189-204.;
Peter T. Martin, Joseph Perrin, Blake Hansen, Ryan Kump, Dan Moore, “Incident Detection Algorithm Evaluation” (University of Utah, Prepared for Utah Department of Transportation) March 2001; and
Dastidar, Samanwoy & Adeli, Hojjat (2003), “Wavelet-Clustering-Neural Network Model for Freeway Incident Detection.”, Computer-Aided Civil and Infrastructure Engineering 18 (5), 325-338.